# Why must the upper limit match when integrating?

1. Nov 23, 2013

### MathewsMD

For example, if you have the function f(x) = x2 then find:

d/dx any number3x∫ t2dt

Why must the dx in d/dx ∫f(t)dt always match the upper limit in order to compute the integral? Why is the lower limit of no concern? I know that you must take chain rule into consideration and change 3x to u, and then do du/dx but why does the upper limit only matter?

Any help would be great!

2. Nov 23, 2013

### Student100

Huh I don't understand your question?

3. Nov 23, 2013

### economicsnerd

When applying the fundamental theorem of calculus, think about everything in terms of differentiation.

Say $f:\mathbb R\to\mathbb R$ is some continuous, bounded function. For any $a,b\in \mathbb R$, let $F(a,b)=\int_a^bf(x)\text{d}x$.

The FTC tells us that differentiating $F(a,b)$ by $b$ gives $f(b)$ (no matter what $a$ is), and differentiating $F(a,b)$ by $a$ gives $-f(a)$ (no matter what $b$ is).

As an example, think about the case where $f(x)>0$ is your speed at time $x$, in which case $F(a,b)$ (for $a<b$) is just how far you've traveled between time $a$ and time $b$.

Talking about how fast you're going at time $b$ doesn't require any information about when you started moving (i.e. time $a$).

4. Nov 23, 2013

### HallsofIvy

Staff Emeritus
You are making some assumptions that are unjustified- in particular that it is the upper
limit that is important!

In order to be able to differentiate with respect to x, you must have a function of x. One way to do that is to have one or the other limit of integration a function of x.
Both $\int_0^x t^2 dt= (1/3)x^3$ and $\int_x^1 e^t dt= e- e^x$ are differentiable with respect to x.

Yet another is to have constant limits of integration while the integrand is a function of both x and the "variable of integration:
$\int_0^1 e^{x+ t}dt= e^x\int_0^1 e^t dt= e^x[e- 1]$ is differentiable with respect to x.

(Strictly speaking, of course, a the integral does NOT have to have an "x" anywhere in order to be differentiable with respect to x! Of course, then, the derivative is 0.)

Last edited by a moderator: Nov 24, 2013
5. Nov 23, 2013

### lurflurf

To elaborate on what Halls said what is important is the places where the variable appears. The general rule is called Leibniz integral rule and states that
$$\dfrac{d}{dt} \int_{\mathrm{a}(t)}^{\mathrm{b}(t)} \mathrm{f}(x,t) \, \mathrm{d}x = \int_{\mathrm{a}(t)}^{\mathrm{b}(t)} \mathrm{f} ^{(0,1)}(x,t) \, \mathrm{d}x + \mathrm{f}(\mathrm{b}(t),t)\mathrm{b}^\prime (t)-\mathrm{f}(\mathrm{a}(t),t)\mathrm{a}^\prime (t) \\ \text{where } \mathrm{f} ^{(0,1)}(x,t) \text{ is the derivative of f with respect to t with x treated as constant.}$$

6. Nov 23, 2013

### economicsnerd

^ This.

And Leibniz's rule is really just what you get by combining FTC and the chain rule.